Research on Reinforcement Learning-Based Dynamic Power Management for Edge Data Center

Guo, Qianying; Huo, Ru; Meng, Hao; Xinhua, E; Liu, Jiang; Huang, Tao

doi:10.1109/icsess.2018.8663880

Cited by 4 publications

(2 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…However, the existing control paradigm is still centralised, where the main computation is conducted at the cloud or control centre [11]. Although many mature intelligent cloud platforms have been in practical use, this operation framework creates dramatic pressures on communication channels because of limited bandwidth [12] - [15]. The 5th generation (5G) mobile network technologies can be a solution, but there are geographical constraints for the equipment at remote locations such as transmission lines and transformers [16].…”

Section: Introductionmentioning

confidence: 99%

Edge-cloud Computing Systems for Smart Grid: State-of-the-art, Architecture, and Applications

Xiang

et al. 2022

Journal of Modern Power Systems and Clean Energy

View full text Add to dashboard Cite

The quantity and heterogeneity of intelligent energy generation and consumption terminals in the smart grid are increasing drastically over the years. These edge devices have created significant pressures on cloud computing (CC) system and centralised control for data storage and processing in realtime operation and control. The integration of edge computing (EC) can effectively alleviate the pressure and conduct real-time processing while ensuring data security. This paper conducts an extensive review of the EC-CC computing system and its application to the smart grid, which will integrate a vast number of dispersed devices. It first comprehensively describes the relationship among CC, fog computing (FC), and EC to provide a theoretical basis for the differentiation. It then introduces the architecture of the EC-CC computing system in the smart grid, where the architecture consists of both hardware structure and software platforms, and key technologies are introduced to support functionalities. Thereafter, the application to the smart grid is discussed across the whole supply chain, including energy generation, transportation (transmission and distribution networks), and consumption. Finally, future research opportunities and challenges of EC-CC while being applied to the smart grid are outlined. This paper can inform future research and industrial exploitations of these new technologies to enable a highly efficient smart grid under decarbonisation, digitalisation, and decentralisation transitions.

show abstract

Section: Introductionmentioning

confidence: 99%

Edge-cloud Computing Systems for Smart Grid: State-of-the-art, Architecture, and Applications

Xiang

et al. 2022

Journal of Modern Power Systems and Clean Energy

View full text Add to dashboard Cite

show abstract

“…Os métodos baseados em aprendizado de máquina podem ser usados para decidir quando alterar o estado do consumo de energia, como é feito em (GUO et al, 2018). A pesquisa utiliza aprendizado por reforço para desenvolver um gerenciador capaz de executar aprendizagem e gerenciamento de energia, de maneira contínua e baseada em eventos.…”

Section: Tipos De Ociosidadeunclassified

Gerenciamento dinâmico de energia aplicado a um robô móvel

Techi¹

View full text Add to dashboard Cite

O presente trabalho foi realizado com apoio da Coordenação de Aperfeiçoamento de Pessoal de Nível Superior -Brasil (CAPES) -Código de Financiamento 001 RESUMOA lei de Moore previu, a cada dois anos, 100% de aumento no número de transistores em um circuito impresso, mantendo o mesmo custo. Baterias têm sua capacidade determinada pela densidade energética (Watt-h/kg), e para que o valor dobrasse foi necessária uma década.Tendo em vista a discrepância entre as taxas de evolução, os equipamentos cada vez consomem mais energia, sem que a fonte consiga armazena-la com o mesmo volume. Tais fatos provocam problemas para sistemas energizados por baterias como, por exemplo, carros elétricos, robôs móveis, casas inteligentes, laptops, smartphones, entre outros. A área da robótica móvel é a mais afetada pois, diferente de carros elétricos, que podem ter grandes bancos de baterias ou de aparelhos eletrônicos que podem ser facilmente utilizados enquanto são recarregados, os sistemas robóticos moveis não possuem espaço suficiente para uma grande quantidade de baterias e sua recarga requer um intervalo de tempo no qual o robô ficara inoperante. Adicionando o fato de um robô móvel ter uma variedade de sensores e atuadores para realizar uma diversa gama de tarefas, o consumo eficiente de energia se torna uma necessidade para proporcionar uma boa experiência para o usuário. O Gerenciamento Dinâmico de Energia (DPM, Dynamic Power Management) propõe o uso otimizado de energia através do controle de gasto energético de equipamentos presentes em um sistema, o que é feito através da exploração da ociosidade: Sempre que um aparelho (ou um componente de um aparelho) estiver ocioso, sua dissipação de energia deve ser reduzida o máximo possível. Este trabalho propõe a modelagem de um sistema de Gerenciamento Dinâmico de Energia em robôs móveis, com objetivo de aumentar o tempo de funcionamento, desenvolver um sistema de captação de dados para futuras tomadas de decisões relacionadas ao sistema de energia e a realização de tarefas, e criar uma base de conhecimento do funcionamento de sensores e atuadores em robôs móveis de plataforma aberta.Palavras-chave: Gerenciamento Dinâmico de Energia. Robô Móvel. Auxilio em decisões voltadas ao consumo de energia.

show abstract

Location-Aware Edge Service Migration for Mobile User Reallocation in Crowded Scenes

Xiao

Yin

Xia

et al. 2021

Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering

View full text Add to dashboard Cite

The mobile edge computing (MEC) paradgim is evolving as an increasingly popular means for developing and deploying smart-cityoriented applications. MEC servers can receive a great deal of requests from equipments of highly mobile users, especially in crowded scenes, e.g., city's central business district (CBD) and school areas. It thus remains a great challenge for appropriate scheduling and managing strategies to avoid hotspots, guarantee load-fairness among MEC servers, and maintain high resource utilization at the same time. To address this challenge, we propose a coalitional-game-based and location-aware approach to MEC Service migration for mobile user reallocation in crowded scenes. Our proposed method includes multiple steps: 1) dividing MEC servers into multiple coalitions according to their inter-euclidean distance by using a modified k-means clustering method; 2) discovering hotspots in every coalition area and scheduling services based on their corresponding cooperations; 3) migrating services to appropriate edge servers to achieve load-fairness among coalition members by using a migration budget mechanism; 4) transferring workloads to nearby coalitions by backbone network in case of workloads beyond the limit. Experimental results based on a real-world mobile trajectory dataset for crowded scenes, and an urban-edge-server-position dataset demonstrate that our method outperforms existing approaches in terms of load-fairness, migration times, and energy consumption of migrations.

show abstract

Research on Reinforcement Learning-Based Dynamic Power Management for Edge Data Center

Cited by 4 publications

References 7 publications

Edge-cloud Computing Systems for Smart Grid: State-of-the-art, Architecture, and Applications

Edge-cloud Computing Systems for Smart Grid: State-of-the-art, Architecture, and Applications

Gerenciamento dinâmico de energia aplicado a um robô móvel

Location-Aware Edge Service Migration for Mobile User Reallocation in Crowded Scenes

Contact Info

Product

Resources

About